104c422 · text · git.iko.soy

git.iko.soy/text

Add a toTitle function

This fixes gh-38.

commit104c42256b718599f1b56bd8e2ceda6df44251a2 authorBryan O'Sullivan <bos@serpentine.com> dateSat 30 Nov 2013, 22:39 (13 years ago) parenta3ac323

3 files changed+84−2

compare with master

unifiedsplit

Data/Text.hs+23 Data/Text/Fusion/Common.hs+37−2 Data/Text/Lazy.hs+24

Data/Text.hs+23

@@ -79,6 +79,7 @@ module Data.Text

7979 , toCaseFold

8080 , toLower

8181 , toUpper

82+ , toTitle

8283

8384 -- ** Justification

8485 , justifyLeft

@@ -679,6 +680,28 @@ toUpper :: Text -> Text

679680toUpper t = unstream (S.toUpper (stream t))

680681{-# INLINE toUpper #-}

681682

683+-- | /O(n)/ Convert a string to title case, using simple case

684+-- conversion.

685+--

686+-- Unlike the other case conversion functions, this function does not

687+-- convert every letter of its input. Instead, the first letter is

688+-- converted to title case, as is every subsequent letter that

689+-- immediately follows a non-letter.

690+--

691+-- The result string may be longer than the input string. For example,

692+-- the Latin small ligature ﬂ (U+FB02) is converted to the

693+-- sequence Latin capital letter F (U+0046) followed by Latin small

694+-- letter l (U+006C).

695+--

696+-- /Note/: this function does not take language or culture specific

697+-- rules into account. For instance, in English, different style

698+-- guides disagree on whether the book name \"The Hill of the Red

699+-- Fox\" is correctly title cased—but this function will

700+-- capitalize /every/ word.

701+toTitle :: Text -> Text

702+toTitle t = unstream (S.toTitle (stream t))

703+{-# INLINE toTitle #-}

704+

682705-- | /O(n)/ Left-justify a string to the given length, using the

683706-- specified fill character on the right. Subject to fusion.

684707-- Performs replacement on invalid scalar values.

Data/Text/Fusion/Common.hs+37−2

@@ -42,6 +42,7 @@ module Data.Text.Fusion.Common

4242 -- $case

4343 , toCaseFold

4444 , toLower

45+ , toTitle

4546 , toUpper

4647

4748 -- ** Justification

@@ -102,13 +103,15 @@ module Data.Text.Fusion.Common

102103

103104import Prelude (Bool(..), Char, Eq(..), Int, Integral, Maybe(..),

104105 Ord(..), Ordering(..), String, (.), ($), (+), (-), (*), (++),

105− (&&), fromIntegral, otherwise)

106+ (&&), fromIntegral, not, otherwise)

106107import qualified Data.List as L

107108import qualified Prelude as P

108109import Data.Bits (shiftL)

110+import Data.Char (isLetter)

109111import Data.Int (Int64)

110112import Data.Text.Fusion.Internal

111−import Data.Text.Fusion.CaseMapping (foldMapping, lowerMapping, upperMapping)

113+import Data.Text.Fusion.CaseMapping (foldMapping, lowerMapping, titleMapping,

114+ upperMapping)

112115import Data.Text.Fusion.Size

113116import GHC.Prim (Addr#, chr#, indexCharOffAddr#, ord#)

114117import GHC.Types (Char(..), Int(..))

@@ -432,6 +435,38 @@ toLower :: Stream Char -> Stream Char

432435toLower = caseConvert lowerMapping

433436{-# INLINE [0] toLower #-}

434437

438+-- | /O(n)/ Convert a string to title case, using simple case

439+-- conversion.

440+--

441+-- Unlike the other case conversion functions, this function does not

442+-- convert every letter of its input. Instead, the first letter is

443+-- converted to title case, as is every subsequent letter that

444+-- immediately follows a non-letter.

445+--

446+-- The result string may be longer than the input string. For example,

447+-- the Latin small ligature ﬂ (U+FB02) is converted to the

448+-- sequence Latin capital letter F (U+0046) followed by Latin small

449+-- letter l (U+006C).

450+--

451+-- /Note/: this function does not take language or culture specific

452+-- rules into account. For instance, in English, different style

453+-- guides disagree on whether the book name \"The Hill of the Red

454+-- Fox\" is correctly title cased—but this function will

455+-- capitalize /every/ word.

456+toTitle :: Stream Char -> Stream Char

457+toTitle (Stream next0 s0 len) = Stream next (CC (False :*: s0) '\0' '\0') len

458+ where

459+ next (CC (letter :*: s) '\0' _) =

460+ case next0 s of

461+ Done -> Done

462+ Skip s' -> Skip (CC (letter :*: s') '\0' '\0')

463+ Yield c s'

464+ | not letter && letter' -> titleMapping c (letter' :*: s')

465+ | otherwise -> Yield c (CC (letter' :*: s') '\0' '\0')

466+ where letter' = isLetter c

467+ next (CC s a b) = Yield a (CC s b '\0')

468+{-# INLINE [0] toTitle #-}

469+

435470justifyLeftI :: Integral a => a -> Char -> Stream Char -> Stream Char

436471justifyLeftI k c (Stream next0 s0 len) =

437472 Stream next (s0 :*: S1 :*: 0) (larger (fromIntegral k) len)

Data/Text/Lazy.hs+24

@@ -86,6 +86,7 @@ module Data.Text.Lazy

8686 , toCaseFold

8787 , toLower

8888 , toUpper

89+ , toTitle

8990

9091 -- ** Justification

9192 , justifyLeft

@@ -697,6 +698,29 @@ toUpper :: Text -> Text

697698toUpper t = unstream (S.toUpper (stream t))

698699{-# INLINE toUpper #-}

699700

701+

702+-- | /O(n)/ Convert a string to title case, using simple case

703+-- conversion.

704+--

705+-- Unlike the other case conversion functions, this function does not

706+-- convert every letter of its input. Instead, the first letter is

707+-- converted to title case, as is every subsequent letter that

708+-- immediately follows a non-letter.

709+--

710+-- The result string may be longer than the input string. For example,

711+-- the Latin small ligature ﬂ (U+FB02) is converted to the

712+-- sequence Latin capital letter F (U+0046) followed by Latin small

713+-- letter l (U+006C).

714+--

715+-- /Note/: this function does not take language or culture specific

716+-- rules into account. For instance, in English, different style

717+-- guides disagree on whether the book name \"The Hill of the Red

718+-- Fox\" is correctly title cased—but this function will

719+-- capitalize /every/ word.

720+toTitle :: Text -> Text

721+toTitle t = unstream (S.toTitle (stream t))

722+{-# INLINE toTitle #-}

723+

700724-- | /O(n)/ 'foldl', applied to a binary operator, a starting value

701725-- (typically the left-identity of the operator), and a 'Text',

702726-- reduces the 'Text' using the binary operator, from left to right.